Star Identification and Attitude Determination with Projective Cameras

Images of starfields collected by a projective camera are useful for a variety of scientific and engineering purposes. This utility is exemplified by star trackers, which are amongst the most commonly used sensors for determining the attitude of modern spacecraft. While the literature on star identification and star-based attitude determination is extensive, most algorithms are developed in an ad hoc manner. This work provides a comprehensive and systematic framework for invariant-based star identification and shows most past star identification algorithms to be special cases within this framework. The new star identification framework is found to motivate new problems in attitude determination and sensor self-calibration. Specifically, new algorithms are presented for simultaneous attitude determination and camera calibration for a generic wide field-of-view sensor using a single starfield image. In the special case where camera focal length is the only unknown calibration parameter, attitude determination performance of the new algorithm is indiscernible from a perfectly calibrated camera.